# Import necessary libraries 

import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
import seaborn as sns
%matplotlib inline

import warnings 
warnings.filterwarnings('ignore')
from sklearn.pipeline import Pipeline
from sklearn.preprocessing import OneHotEncoder,StandardScaler,PowerTransformer
from sklearn.model_selection import train_test_split
from sklearn.feature_selection import RFE
from sklearn import metrics
from sklearn.metrics import precision_recall_curve

import statsmodels.api as sm
from statsmodels.stats.outliers_influence import variance_inflation_factor

# Load and Read the dataset
df = pd.read_csv("Leads.csv")
df.head()

# check the shape
df.shape

(9240, 37)

df_copy = df.copy()

df.info()

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 9240 entries, 0 to 9239
Data columns (total 37 columns):
 #   Column                                         Non-Null Count  Dtype  
---  ------                                         --------------  -----  
 0   Prospect ID                                    9240 non-null   object 
 1   Lead Number                                    9240 non-null   int64  
 2   Lead Origin                                    9240 non-null   object 
 3   Lead Source                                    9204 non-null   object 
 4   Do Not Email                                   9240 non-null   object 
 5   Do Not Call                                    9240 non-null   object 
 6   Converted                                      9240 non-null   int64  
 7   TotalVisits                                    9103 non-null   float64
 8   Total Time Spent on Website                    9240 non-null   int64  
 9   Page Views Per Visit                           9103 non-null   float64
 10  Last Activity                                  9137 non-null   object 
 11  Country                                        6779 non-null   object 
 12  Specialization                                 7802 non-null   object 
 13  How did you hear about X Education             7033 non-null   object 
 14  What is your current occupation                6550 non-null   object 
 15  What matters most to you in choosing a course  6531 non-null   object 
 16  Search                                         9240 non-null   object 
 17  Magazine                                       9240 non-null   object 
 18  Newspaper Article                              9240 non-null   object 
 19  X Education Forums                             9240 non-null   object 
 20  Newspaper                                      9240 non-null   object 
 21  Digital Advertisement                          9240 non-null   object 
 22  Through Recommendations                        9240 non-null   object 
 23  Receive More Updates About Our Courses         9240 non-null   object 
 24  Tags                                           5887 non-null   object 
 25  Lead Quality                                   4473 non-null   object 
 26  Update me on Supply Chain Content              9240 non-null   object 
 27  Get updates on DM Content                      9240 non-null   object 
 28  Lead Profile                                   6531 non-null   object 
 29  City                                           7820 non-null   object 
 30  Asymmetrique Activity Index                    5022 non-null   object 
 31  Asymmetrique Profile Index                     5022 non-null   object 
 32  Asymmetrique Activity Score                    5022 non-null   float64
 33  Asymmetrique Profile Score                     5022 non-null   float64
 34  I agree to pay the amount through cheque       9240 non-null   object 
 35  A free copy of Mastering The Interview         9240 non-null   object 
 36  Last Notable Activity                          9240 non-null   object 
dtypes: float64(4), int64(3), object(30)
memory usage: 2.6+ MB

df.describe()

# Remove duplicates

df.drop_duplicates(inplace=True) # inplace = True to make the changes permanent

# Check for the null values using isnull() function

df.isnull().sum()

Prospect ID                                         0
Lead Number                                         0
Lead Origin                                         0
Lead Source                                        36
Do Not Email                                        0
Do Not Call                                         0
Converted                                           0
TotalVisits                                       137
Total Time Spent on Website                         0
Page Views Per Visit                              137
Last Activity                                     103
Country                                          2461
Specialization                                   1438
How did you hear about X Education               2207
What is your current occupation                  2690
What matters most to you in choosing a course    2709
Search                                              0
Magazine                                            0
Newspaper Article                                   0
X Education Forums                                  0
Newspaper                                           0
Digital Advertisement                               0
Through Recommendations                             0
Receive More Updates About Our Courses              0
Tags                                             3353
Lead Quality                                     4767
Update me on Supply Chain Content                   0
Get updates on DM Content                           0
Lead Profile                                     2709
City                                             1420
Asymmetrique Activity Index                      4218
Asymmetrique Profile Index                       4218
Asymmetrique Activity Score                      4218
Asymmetrique Profile Score                       4218
I agree to pay the amount through cheque            0
A free copy of Mastering The Interview              0
Last Notable Activity                               0
dtype: int64

# converting null values into percentages for better understanding
round((df.isnull().sum()/len(df)*100),2)

Prospect ID                                       0.00
Lead Number                                       0.00
Lead Origin                                       0.00
Lead Source                                       0.39
Do Not Email                                      0.00
Do Not Call                                       0.00
Converted                                         0.00
TotalVisits                                       1.48
Total Time Spent on Website                       0.00
Page Views Per Visit                              1.48
Last Activity                                     1.11
Country                                          26.63
Specialization                                   15.56
How did you hear about X Education               23.89
What is your current occupation                  29.11
What matters most to you in choosing a course    29.32
Search                                            0.00
Magazine                                          0.00
Newspaper Article                                 0.00
X Education Forums                                0.00
Newspaper                                         0.00
Digital Advertisement                             0.00
Through Recommendations                           0.00
Receive More Updates About Our Courses            0.00
Tags                                             36.29
Lead Quality                                     51.59
Update me on Supply Chain Content                 0.00
Get updates on DM Content                         0.00
Lead Profile                                     29.32
City                                             15.37
Asymmetrique Activity Index                      45.65
Asymmetrique Profile Index                       45.65
Asymmetrique Activity Score                      45.65
Asymmetrique Profile Score                       45.65
I agree to pay the amount through cheque          0.00
A free copy of Mastering The Interview            0.00
Last Notable Activity                             0.00
dtype: float64

# Handle 'SELECT' or placeholder values ( treat them as NaN)

df.replace('Select',np.nan, inplace=True)

# Drop columns with too many missing values [missing values >40%]

missing_ratio = df.isnull().mean()
cols_to_drop = missing_ratio[missing_ratio > 0.4].index
df.drop(columns = cols_to_drop,inplace=True)

df.shape

(9240, 30)

# Handling null values for categorical columns (replacing with mode )
for col in df.select_dtypes(include='object').columns :
    df[col].fillna(df[col].mode()[0],inplace = True)

# Handling null values for numerical columns ( replacing with the median )

for col in df.select_dtypes(include =['int64','float64']).columns:
    df[col].fillna(df[col].median(),inplace = True)

df.isnull().sum()

Prospect ID                                      0
Lead Number                                      0
Lead Origin                                      0
Lead Source                                      0
Do Not Email                                     0
Do Not Call                                      0
Converted                                        0
TotalVisits                                      0
Total Time Spent on Website                      0
Page Views Per Visit                             0
Last Activity                                    0
Country                                          0
Specialization                                   0
What is your current occupation                  0
What matters most to you in choosing a course    0
Search                                           0
Magazine                                         0
Newspaper Article                                0
X Education Forums                               0
Newspaper                                        0
Digital Advertisement                            0
Through Recommendations                          0
Receive More Updates About Our Courses           0
Tags                                             0
Update me on Supply Chain Content                0
Get updates on DM Content                        0
City                                             0
I agree to pay the amount through cheque         0
A free copy of Mastering The Interview           0
Last Notable Activity                            0
dtype: int64

# check the outliers present in the data set by check the quantilies [.25, .5, .75, .90, .95, .99]
df.describe(percentiles=[.25, .5, .75, .90, .95, .99])

# Visualize the outliers in the numerical columns
num_col = df.select_dtypes(include=['int64','float64']).columns
for i in num_col:
    if i != 'Converted':
        plt.figure(figsize = (6,4))
        sns.boxplot(df[i])
        plt.show()
    
#sns.boxplot(df['Lead Number'])

# Checking for outliers and Handling them using Inter Quartile Range

num_col = df.select_dtypes(include=['int64','float64']).columns

for i in num_col:
    Q1 = df[i].quantile(0.25)
    Q3 = df[i].quantile(0.75)
    IQR = Q3-Q1
    lower_bound = Q1-1.5*IQR
    upper_bound = Q3+1.5*IQR
    df[i] = np.where(df[i] > upper_bound, upper_bound,
                      np.where(df[i] < lower_bound, lower_bound, df[i])) # Capping the outliers

df.shape

(9240, 30)

# Visualize the outliers in the numerical columns after handling them
num_col = df.select_dtypes(include=['int64','float64']).columns
for i in num_col:
    if i != 'Converted':
        plt.figure(figsize = (6,4))
        sns.boxplot(df[i])
        plt.show()
    
print("Outliers of the numerical columns are handled")

Outliers of the numerical columns are handled

numerical_col = df.select_dtypes(include = ['int64','float64']).columns
categorical_col = df.select_dtypes(include = ['object']).columns
numerical_col

Index(['Lead Number', 'Converted', 'TotalVisits',
       'Total Time Spent on Website', 'Page Views Per Visit'],
      dtype='object')

df.head()

variable_list = ['Do Not Email', 'Do Not Call', 'Search', 'Magazine', 'Newspaper Article', 
               'X Education Forums', 'Newspaper', 'Digital Advertisement', 
               'Through Recommendations', 'Receive More Updates About Our Courses',
               'I agree to pay the amount through cheque', 'A free copy of Mastering The Interview','Update me on Supply Chain Content','Get updates on DM Content']

#df[variable_list] = df[variable_list].apply(lambda x: x.map({'Yes': 1,'No': 0}))
# creating a function binary_map to convert yes/no columns to 0/1

def binary_map(x):
    return x.map({'Yes': 1, "No": 0})

# Applying the function to the housing list
df[variable_list] = df[variable_list].apply(binary_map)

df.head()

# Remove the column 'Prospect ID' it is of no use to us
df.drop('Prospect ID',axis =1, inplace = True)

categorical_cols = df.select_dtypes(include='object').columns.tolist()

categorical_cols

['Lead Origin',
 'Lead Source',
 'Last Activity',
 'Country',
 'Specialization',
 'What is your current occupation',
 'What matters most to you in choosing a course',
 'Tags',
 'City',
 'Last Notable Activity']

# create dummy variables for the categorical columns
dummy_variables = pd.get_dummies(df[categorical_cols], drop_first= True).astype(int)

# Merge the dummy variables with the original dataset
df = pd.concat([df,dummy_variables],axis=1)
df.shape

(9240, 175)

# Drop the categorical_cols from the data as we have created dummy variables for them
df.drop(categorical_cols, axis =1,inplace = True)

df.shape

(9240, 165)

df.head()

lead_conversion_rate = (sum(df['Converted'])/len(df['Converted'].index))*100
lead_conversion_rate

38.53896103896104

# Correlation matrix using heatmap
plt.figure(figsize=(60,40))
sns.heatmap(df.corr(),annot = True)
plt.show()

# Heatmap for numerical columns
plt.figure(figsize = (12,10))
sns.heatmap(df.select_dtypes(include=['int64','int32']).corr(), annot = True)
plt.show()

# heatmap for only float value columns
plt.figure(figsize = (12,10))
sns.heatmap(df.select_dtypes(include=['float']).corr(), annot = True)
plt.show()

# Create correlation matrix for dummy_variables
leads_encoded = pd.get_dummies(df, drop_first=True)

plt.figure(figsize=(20,10))
sns.heatmap(leads_encoded.corr(), annot=False, cmap="coolwarm")  # annot=False for clarity
plt.show()

# correlation matrix using top features
corr_matrix = leads_encoded.corr()

# pick top 20 features most correlated with Converted
top_features = corr_matrix["Converted"].abs().sort_values(ascending=False).head(20).index

plt.figure(figsize=(12,8))
sns.heatmap(corr_matrix.loc[top_features, top_features], annot=True, fmt=".2f")
plt.title("Top 20 Correlated Features")
plt.show()

# dropping the columns which are highly correlated
df.drop(['What is your current occupation_Unemployed', 'What is your current occupation_Working Professional'], axis=1, inplace=True)

# Get new dummy variables after dropping the higly correlated columns
leads_encoded_new = pd.get_dummies(df, drop_first=True)

plt.figure(figsize=(20,10))
sns.heatmap(leads_encoded.corr(), annot=False, cmap="coolwarm")  # annot=False for clarity
plt.show()

# correlation matrix
corr_matrix = leads_encoded_new.corr()
# pick top 20 features most correlated with Converted
top_features = corr_matrix["Converted"].abs().sort_values(ascending=False).head(20).index

plt.figure(figsize=(12,8))
sns.heatmap(corr_matrix.loc[top_features, top_features], annot=True, fmt=".2f")
plt.title("Top 20 Correlated Features")
plt.show()

# splitting the data into input and target variables
x = df.drop(columns =['Converted','Lead Number'],axis=1)
y = df['Converted']
x.shape

(9240, 161)

# Let's start with splitting data into train and test data

x_train,x_test,y_train,y_test = train_test_split(x,y, test_size=0.2,random_state= 42)

# Feature scaling using standard scaler
scaler = StandardScaler()
x_train[['TotalVisits','Total Time Spent on Website','Page Views Per Visit']]= scaler.fit_transform(x_train[['TotalVisits',
                                                                                                             'Total Time Spent on Website',
                                                                                                             'Page Views Per Visit']])

x_train.head()

# Insight 1: Overall Conversion Rate
import matplotlib.pyplot as plt
import seaborn as sns
conversion_rate = y.mean() * 100
plt.figure(figsize=(5, 4))
y.value_counts().plot(kind='bar')
plt.title(f'Lead Conversion Distribution (Rate: {conversion_rate:.2f}%)')
plt.xlabel('Converted')
plt.ylabel('Count')
plt.show()
print(f"INSIGHT 1: Overall conversion rate is {conversion_rate:.2f}%, a benchmark for lead nurturing success.")

INSIGHT 1: Overall conversion rate is 38.54%, a benchmark for lead nurturing success.

# Insight 2: Conversion by Lead Source
lead_source_conv = df_copy.groupby('Lead Source')['Converted'].mean().sort_values(ascending=False)
plt.figure(figsize=(10, 4))
lead_source_conv.plot(kind='bar')
plt.title('Conversion Rate by Lead Source')
plt.ylabel('Conversion Rate')
plt.xticks(rotation=45, ha='right')
plt.show()
print("INSIGHT 2: High-converting sources should get priority in marketing budgets.")

INSIGHT 2: High-converting sources should get priority in marketing budgets.

# Insight 3: Conversion by Lead Origin
lead_origin_conv = df_copy.groupby('Lead Origin')['Converted'].mean().sort_values(ascending=False)
plt.figure(figsize=(8, 4))
lead_origin_conv.plot(kind='bar')
plt.title('Conversion Rate by Lead Origin')
plt.ylabel('Conversion Rate')
plt.xticks(rotation=45, ha='right')
plt.show()
print("INSIGHT 3: Lead Origin is a strong segmentation variable — optimize acquisition channels accordingly.")

INSIGHT 3: Lead Origin is a strong segmentation variable — optimize acquisition channels accordingly.

# Insight 4: Total Visits vs Conversion
if 'TotalVisits' in x.columns:
    df_visits = df.copy()
    df_visits['TotalVisits_bin'] = pd.cut(df_visits['TotalVisits'], bins=[0, 1, 3, 5, 10, np.inf], labels=['0-1', '2-3', '4-5', '6-10', '10+'])
    visit_conv = df_visits.groupby('TotalVisits_bin')['Converted'].mean()
    plt.figure(figsize=(7, 4))
    visit_conv.plot(kind='bar')
    plt.title('Conversion Rate by Total Visits')
    plt.ylabel('Conversion Rate')
    plt.show()
    print("INSIGHT 4: Visitors with more site visits have higher conversion probability — retarget low-visit users.")

INSIGHT 4: Visitors with more site visits have higher conversion probability — retarget low-visit users.

# Insight 5: Total Time Spent vs Conversion
if 'Total Time Spent on Website' in x.columns:
    plt.figure(figsize=(8, 4))
    df.boxplot(column='Total Time Spent on Website', by='Converted')
    plt.suptitle('')
    plt.title('Website Time vs Conversion')
    plt.ylabel('Total Time Spent')
    plt.show()
    print("INSIGHT 5: Higher engagement time strongly correlates with conversion.")

<Figure size 800x400 with 0 Axes>

INSIGHT 5: Higher engagement time strongly correlates with conversion.

import statsmodels.api as sm

# Logistic Regression Model
logm1 = sm.GLM(y_train, (sm.add_constant(x_train)), family = sm.families.Binomial())
logm1.fit().summary()

from sklearn.linear_model import LogisticRegression
logreg = LogisticRegression(max_iter=1000)

from sklearn.feature_selection import RFE
rfe = RFE(logreg, n_features_to_select=15)
rfe = rfe.fit(x_train,y_train)

rfe.support_

array([ True, False, False, False, False, False, False, False, False,
       False, False, False, False, False, False, False, False, False,
        True, False, False, False, False, False, False, False, False,
       False, False, False, False, False, False, False,  True, False,
       False, False, False, False, False, False, False, False, False,
       False, False, False, False, False, False, False, False, False,
       False, False, False, False, False, False, False, False, False,
       False, False, False, False, False, False, False, False, False,
       False, False, False, False, False, False, False, False, False,
       False, False, False, False, False, False, False, False, False,
       False, False, False, False, False, False, False, False, False,
       False, False, False, False, False, False, False, False, False,
       False, False, False, False, False, False, False, False,  True,
        True, False, False, False, False, False, False, False,  True,
       False, False, False,  True, False, False, False, False,  True,
       False,  True,  True, False,  True,  True, False, False, False,
       False, False, False, False, False, False, False, False,  True,
       False, False, False, False,  True, False,  True, False])

list(zip(x_train.columns, rfe.support_, rfe.ranking_))

[('Do Not Email', True, 1),
 ('Do Not Call', False, 99),
 ('TotalVisits', False, 81),
 ('Total Time Spent on Website', False, 2),
 ('Page Views Per Visit', False, 80),
 ('Search', False, 110),
 ('Magazine', False, 134),
 ('Newspaper Article', False, 65),
 ('X Education Forums', False, 64),
 ('Newspaper', False, 118),
 ('Digital Advertisement', False, 79),
 ('Through Recommendations', False, 49),
 ('Receive More Updates About Our Courses', False, 144),
 ('Update me on Supply Chain Content', False, 146),
 ('Get updates on DM Content', False, 141),
 ('I agree to pay the amount through cheque', False, 136),
 ('A free copy of Mastering The Interview', False, 57),
 ('Lead Origin_Landing Page Submission', False, 9),
 ('Lead Origin_Lead Add Form', True, 1),
 ('Lead Origin_Lead Import', False, 26),
 ('Lead Origin_Quick Add Form', False, 73),
 ('Lead Source_Direct Traffic', False, 78),
 ('Lead Source_Facebook', False, 100),
 ('Lead Source_Google', False, 85),
 ('Lead Source_Live Chat', False, 114),
 ('Lead Source_NC_EDM', False, 67),
 ('Lead Source_Olark Chat', False, 27),
 ('Lead Source_Organic Search', False, 101),
 ('Lead Source_Pay per Click Ads', False, 127),
 ('Lead Source_Press_Release', False, 142),
 ('Lead Source_Reference', False, 40),
 ('Lead Source_Referral Sites', False, 84),
 ('Lead Source_Social Media', False, 37),
 ('Lead Source_WeLearn', False, 105),
 ('Lead Source_Welingak Website', True, 1),
 ('Lead Source_bing', False, 53),
 ('Lead Source_blog', False, 45),
 ('Lead Source_google', False, 34),
 ('Lead Source_testone', False, 123),
 ('Lead Source_welearnblog_Home', False, 140),
 ('Lead Source_youtubechannel', False, 56),
 ('Last Activity_Converted to Lead', False, 47),
 ('Last Activity_Email Bounced', False, 21),
 ('Last Activity_Email Link Clicked', False, 108),
 ('Last Activity_Email Marked Spam', False, 66),
 ('Last Activity_Email Opened', False, 121),
 ('Last Activity_Email Received', False, 29),
 ('Last Activity_Form Submitted on Website', False, 111),
 ('Last Activity_Had a Phone Conversation', False, 12),
 ('Last Activity_Olark Chat Conversation', False, 24),
 ('Last Activity_Page Visited on Website', False, 63),
 ('Last Activity_Resubscribed to emails', False, 138),
 ('Last Activity_SMS Sent', False, 14),
 ('Last Activity_Unreachable', False, 106),
 ('Last Activity_Unsubscribed', False, 82),
 ('Last Activity_View in browser link Clicked', False, 86),
 ('Last Activity_Visited Booth in Tradeshow', False, 133),
 ('Country_Australia', False, 116),
 ('Country_Bahrain', False, 74),
 ('Country_Bangladesh', False, 44),
 ('Country_Belgium', False, 41),
 ('Country_Canada', False, 35),
 ('Country_China', False, 88),
 ('Country_Denmark', False, 122),
 ('Country_France', False, 87),
 ('Country_Germany', False, 28),
 ('Country_Ghana', False, 75),
 ('Country_Hong Kong', False, 30),
 ('Country_India', False, 72),
 ('Country_Indonesia', False, 143),
 ('Country_Italy', False, 11),
 ('Country_Kenya', False, 129),
 ('Country_Kuwait', False, 95),
 ('Country_Liberia', False, 115),
 ('Country_Malaysia', False, 126),
 ('Country_Netherlands', False, 102),
 ('Country_Nigeria', False, 76),
 ('Country_Oman', False, 96),
 ('Country_Philippines', False, 128),
 ('Country_Qatar', False, 61),
 ('Country_Russia', False, 147),
 ('Country_Saudi Arabia', False, 22),
 ('Country_Singapore', False, 42),
 ('Country_South Africa', False, 125),
 ('Country_Sri Lanka', False, 132),
 ('Country_Sweden', False, 107),
 ('Country_Switzerland', False, 46),
 ('Country_Tanzania', False, 131),
 ('Country_Uganda', False, 120),
 ('Country_United Arab Emirates', False, 69),
 ('Country_United Kingdom', False, 18),
 ('Country_United States', False, 36),
 ('Country_Vietnam', False, 135),
 ('Country_unknown', False, 117),
 ('Specialization_Business Administration', False, 62),
 ('Specialization_E-Business', False, 48),
 ('Specialization_E-COMMERCE', False, 70),
 ('Specialization_Finance Management', False, 8),
 ('Specialization_Healthcare Management', False, 93),
 ('Specialization_Hospitality Management', False, 55),
 ('Specialization_Human Resource Management', False, 92),
 ('Specialization_IT Projects Management', False, 112),
 ('Specialization_International Business', False, 43),
 ('Specialization_Marketing Management', False, 130),
 ('Specialization_Media and Advertising', False, 90),
 ('Specialization_Operations Management', False, 103),
 ('Specialization_Retail Management', False, 50),
 ('Specialization_Rural and Agribusiness', False, 104),
 ('Specialization_Services Excellence', False, 91),
 ('Specialization_Supply Chain Management', False, 89),
 ('Specialization_Travel and Tourism', False, 94),
 ('What is your current occupation_Housewife', False, 4),
 ('What is your current occupation_Other', False, 17),
 ('What is your current occupation_Student', False, 13),
 ('What matters most to you in choosing a course_Flexibility & Convenience',
  False,
  113),
 ('What matters most to you in choosing a course_Other', False, 124),
 ('Tags_Busy', True, 1),
 ('Tags_Closed by Horizzon', True, 1),
 ('Tags_Diploma holder (Not Eligible)', False, 19),
 ('Tags_Graduation in progress', False, 32),
 ('Tags_In confusion whether part time or DLP', False, 16),
 ('Tags_Interested  in full time MBA', False, 52),
 ('Tags_Interested in Next batch', False, 10),
 ('Tags_Interested in other courses', False, 38),
 ('Tags_Lateral student', False, 15),
 ('Tags_Lost to EINS', True, 1),
 ('Tags_Lost to Others', False, 54),
 ('Tags_Not doing further education', False, 31),
 ('Tags_Recognition issue (DEC approval)', False, 145),
 ('Tags_Ringing', True, 1),
 ('Tags_Shall take in the next coming month', False, 97),
 ('Tags_Still Thinking', False, 51),
 ('Tags_University not recognized', False, 109),
 ('Tags_Want to take admission but has financial problems', False, 23),
 ('Tags_Will revert after reading the email', True, 1),
 ('Tags_in touch with EINS', False, 5),
 ('Tags_invalid number', True, 1),
 ('Tags_number not provided', True, 1),
 ('Tags_opp hangup', False, 60),
 ('Tags_switched off', True, 1),
 ('Tags_wrong number given', True, 1),
 ('City_Other Cities', False, 59),
 ('City_Other Cities of Maharashtra', False, 98),
 ('City_Other Metro Cities', False, 77),
 ('City_Thane & Outskirts', False, 68),
 ('City_Tier II Cities', False, 58),
 ('Last Notable Activity_Email Bounced', False, 20),
 ('Last Notable Activity_Email Link Clicked', False, 25),
 ('Last Notable Activity_Email Marked Spam', False, 71),
 ('Last Notable Activity_Email Opened', False, 39),
 ('Last Notable Activity_Email Received', False, 83),
 ('Last Notable Activity_Form Submitted on Website', False, 137),
 ('Last Notable Activity_Had a Phone Conversation', True, 1),
 ('Last Notable Activity_Modified', False, 6),
 ('Last Notable Activity_Olark Chat Conversation', False, 3),
 ('Last Notable Activity_Page Visited on Website', False, 33),
 ('Last Notable Activity_Resubscribed to emails', False, 139),
 ('Last Notable Activity_SMS Sent', True, 1),
 ('Last Notable Activity_Unreachable', False, 7),
 ('Last Notable Activity_Unsubscribed', True, 1),
 ('Last Notable Activity_View in browser link Clicked', False, 119)]

# columns which are selected from RFE model
col = x_train.columns[rfe.support_]
col

Index(['Do Not Email', 'Lead Origin_Lead Add Form',
       'Lead Source_Welingak Website', 'Tags_Busy', 'Tags_Closed by Horizzon',
       'Tags_Lost to EINS', 'Tags_Ringing',
       'Tags_Will revert after reading the email', 'Tags_invalid number',
       'Tags_number not provided', 'Tags_switched off',
       'Tags_wrong number given',
       'Last Notable Activity_Had a Phone Conversation',
       'Last Notable Activity_SMS Sent', 'Last Notable Activity_Unsubscribed'],
      dtype='object')

x_train_sm = sm.add_constant(x_train[col])
logm2 = sm.GLM(y_train,x_train_sm, sm.families.Binomial())
res = logm2.fit()
res.summary()

# Make predictions on the training data
y_train_pred = res.predict(x_train_sm)
y_train_pred

6487    0.015654
4759    0.357183
4368    0.024448
1467    0.901694
5517    0.024448
          ...   
5734    0.004408
5191    0.357183
5390    0.990350
860     0.004408
7270    0.357183
Length: 7392, dtype: float64

# Reshape values 
y_train_pred = y_train_pred.values.reshape(-1)
y_train_pred[:10]

array([0.01565423, 0.35718324, 0.02444791, 0.90169386, 0.02444791,
       0.00440771, 0.35718324, 0.35718324, 0.97155985, 0.02444791])

y_train_pred_final = pd.DataFrame({'Converted':y_train.values,'converted_prob':y_train_pred})
y_train_pred_final['Prospect ID'] = y_train.index
y_train_pred_final

y_train_pred_final['Predicted'] = y_train_pred_final.converted_prob.map(lambda x : 1 if x>0.5 else 0)
y_train_pred_final

from sklearn import metrics
confusion_matrix = metrics.confusion_matrix(y_train_pred_final.Converted,y_train_pred_final.Predicted)
confusion_matrix

array([[4428,  144],
       [1060, 1760]], dtype=int64)

accuracy_score = metrics.accuracy_score(y_train_pred_final.Converted,y_train_pred_final.Predicted)
accuracy_score

0.8371212121212122

from statsmodels.stats.outliers_influence import variance_inflation_factor

vif = pd.DataFrame()
vif['Features'] = x_train[col].columns
vif['VIF'] = [variance_inflation_factor(x_train[col].values,i) for i in range(x_train[col].shape[1])]
vif['VIF'] = round(vif['VIF'],2)
vif = vif.sort_values(by = 'VIF',ascending = False)
vif

col = col.drop('Tags_number not provided',1)

x_train_sm = sm.add_constant(x_train[col])
logm3 = sm.GLM(y_train,x_train_sm, sm.families.Binomial())
res = logm3.fit()
res.summary()

y_train_pred = res.predict(x_train_sm)
y_train_pred

6487    0.015583
4759    0.357612
4368    0.022982
1467    0.900540
5517    0.022982
          ...   
5734    0.004461
5191    0.357612
5390    0.990270
860     0.004461
7270    0.357612
Length: 7392, dtype: float64

y_train_pred = y_train_pred.values.reshape(-1)
y_train_pred[:10]

array([0.01558295, 0.35761196, 0.02298228, 0.90053976, 0.02298228,
       0.00446149, 0.35761196, 0.35761196, 0.97156333, 0.02298228])

# y_train_pred_final = pd.DataFrame({'Converted':y_train.values,'converted_prob':y_train_pred})
# y_train_pred_final['Prospect ID'] = y_train.index
# y_train_pred_final

y_train_pred_final['converted_prob'] = y_train_pred

y_train_pred_final['Predicted'] = y_train_pred_final.converted_prob.map(lambda x : 1 if x>0.5 else 0)
y_train_pred_final

from sklearn import metrics
confusion_matrix2 = metrics.confusion_matrix(y_train_pred_final.Converted,y_train_pred_final.Predicted)
confusion_matrix2

array([[4428,  144],
       [1061, 1759]], dtype=int64)

accuracy_score = metrics.accuracy_score(y_train_pred_final.Converted,y_train_pred_final.Predicted)
accuracy_score

0.8369859307359307

from statsmodels.stats.outliers_influence import variance_inflation_factor

vif = pd.DataFrame()
vif['Features'] = x_train[col].columns
vif['VIF'] = [variance_inflation_factor(x_train[col].values,i) for i in range(x_train[col].shape[1])]
vif['VIF'] = round(vif['VIF'],2)
vif = vif.sort_values(by = 'VIF',ascending = False)
vif

# Let's drop the column 'Tags_wrong number given' which has high p-value
col = col.drop('Tags_wrong number given')
col

Index(['Do Not Email', 'Lead Origin_Lead Add Form',
       'Lead Source_Welingak Website', 'Tags_Busy', 'Tags_Closed by Horizzon',
       'Tags_Lost to EINS', 'Tags_Ringing',
       'Tags_Will revert after reading the email', 'Tags_invalid number',
       'Tags_switched off', 'Last Notable Activity_Had a Phone Conversation',
       'Last Notable Activity_SMS Sent', 'Last Notable Activity_Unsubscribed'],
      dtype='object')

x_train_sm = sm.add_constant(x_train[col])
logm4 = sm.GLM(y_train,x_train_sm, sm.families.Binomial())
res = logm4.fit()
res.summary()

y_train_pred = res.predict(x_train_sm).values.reshape(-1)
y_train_pred[:10]

array([0.01540345, 0.35820124, 0.02107786, 0.89911781, 0.02107786,
       0.00452884, 0.35820124, 0.35820124, 0.97158096, 0.02107786])

y_train_pred_final['converted_prob'] = y_train_pred

# Creating new column 'predicted' with 1 if Churn_Prob > 0.5 else 0
y_train_pred_final['Predicted'] = y_train_pred_final.converted_prob.map(lambda x: 1 if x > 0.5 else 0)
y_train_pred_final.head()

confusion_matrix3 = metrics.confusion_matrix(y_train_pred_final.Converted,y_train_pred_final.Predicted)
confusion_matrix3

array([[4428,  144],
       [1061, 1759]], dtype=int64)

# Let's check the overall accuracy.
print(metrics.accuracy_score(y_train_pred_final.Converted, y_train_pred_final.Predicted))

0.8369859307359307

vif = pd.DataFrame()
vif['Features'] = x_train[col].columns
vif['VIF'] = [variance_inflation_factor(x_train[col].values,i) for i in range(x_train[col].shape[1])]
vif['VIF'] = round(vif['VIF'],2)
vif = vif.sort_values(by = 'VIF',ascending = False)
vif

# Let's drop the column 'Tags_wrong number given' which has high p-value
col = col.drop('Tags_invalid number')
col

Index(['Do Not Email', 'Lead Origin_Lead Add Form',
       'Lead Source_Welingak Website', 'Tags_Busy', 'Tags_Closed by Horizzon',
       'Tags_Lost to EINS', 'Tags_Ringing',
       'Tags_Will revert after reading the email', 'Tags_switched off',
       'Last Notable Activity_Had a Phone Conversation',
       'Last Notable Activity_SMS Sent', 'Last Notable Activity_Unsubscribed'],
      dtype='object')

x_train_sm = sm.add_constant(x_train[col])
logm5 = sm.GLM(y_train,x_train_sm, sm.families.Binomial())
res = logm5.fit()
res.summary()

y_train_pred = res.predict(x_train_sm).values.reshape(-1)
y_train_pred[:10]

array([0.01540822, 0.35877541, 0.01880419, 0.89737471, 0.01880419,
       0.00460772, 0.35877541, 0.35877541, 0.97156911, 0.01880419])

y_train_pred_final['converted_prob'] = y_train_pred

# Creating new column 'predicted' with 1 if Churn_Prob > 0.5 else 0
y_train_pred_final['Predicted'] = y_train_pred_final.converted_prob.map(lambda x: 1 if x > 0.5 else 0)
y_train_pred_final.head()

confusion_matrix4 = metrics.confusion_matrix(y_train_pred_final.Converted,y_train_pred_final.Predicted)
confusion_matrix4

array([[4435,  137],
       [1069, 1751]], dtype=int64)

# Let's check the overall accuracy.
print(metrics.accuracy_score(y_train_pred_final.Converted, y_train_pred_final.Predicted))

0.8368506493506493

vif = pd.DataFrame()
vif['Features'] = x_train[col].columns
vif['VIF'] = [variance_inflation_factor(x_train[col].values,i) for i in range(x_train[col].shape[1])]
vif['VIF'] = round(vif['VIF'],2)
vif = vif.sort_values(by = 'VIF',ascending = False)
vif

confusion_matrix5 = metrics.confusion_matrix(y_train_pred_final.Converted,y_train_pred_final.Predicted)
confusion_matrix5

array([[4435,  137],
       [1069, 1751]], dtype=int64)

TP = confusion_matrix[1,1]  # True positive
FP = confusion_matrix[0,1]  # False positive
TN = confusion_matrix[0,0]  # True Negative
FN = confusion_matrix[1,0]  # False Negative

# Let's find out the sesitivity of the model

 TP/float(TP+FN)

0.624113475177305

# let's findout the specificity
TN/float(TN+FP)

0.968503937007874

# Calculate false postive rate - predicting churn when customer does not have churned
FP/float(FP+TN)

0.031496062992125984

# positive predictive rate
TP/float(TP+FP)

0.9243697478991597

# negative predictive rate
TN/float(TN+FN)

0.8068513119533528

def draw_roc( actual, probs ):
    fpr, tpr, thresholds = metrics.roc_curve( actual, probs,
                                              drop_intermediate = False )
    auc_score = metrics.roc_auc_score( actual, probs )
    plt.figure(figsize=(5, 5))
    plt.plot( fpr, tpr, label='ROC curve (area = %0.2f)' % auc_score )
    plt.plot([0, 1], [0, 1], 'k--')
    plt.xlim([0.0, 1.0])
    plt.ylim([0.0, 1.05])
    plt.xlabel('False Positive Rate or [1 - True Negative Rate]')
    plt.ylabel('True Positive Rate')
    plt.title('Receiver operating characteristic example')
    plt.legend(loc="lower right")
    plt.show()

    return None

fpr, tpr, thresholds = metrics.roc_curve( y_train_pred_final.Converted, y_train_pred_final.converted_prob, drop_intermediate = False )

draw_roc(y_train_pred_final.Converted, y_train_pred_final.converted_prob)

# let's create columns with different cuttoff points

numbers = [float(x)/10 for x in range(10)]

for i in numbers:
    y_train_pred_final[i] = y_train_pred_final.converted_prob.map(lambda x : 1 if x>i else 0)

y_train_pred_final.head()

# Now lets calculate Accuracy, sensitivity and  Specificity for various cutoff probabalities

cutoff_df = pd.DataFrame(columns = ['prob','Accuracy','sensitivity','specificity'])

num = [0.0,0.1,0.2,0.3,0.4,0.5,0.6,0.7,0.8,0.9]
for i in num:
    cm1 = metrics.confusion_matrix(y_train_pred_final.Converted,y_train_pred_final[i])
    total1=sum(sum(cm1))
    Accuracy = (cm1[0,0]+cm1[1,1])/total1
    specificity = cm1[0,0]/(cm1[0,0]+cm1[0,1])
    sensitivity = cm1[1,1]/(cm1[1,0]+cm1[1,1])

    cutoff_df.loc[i] = [i, Accuracy, sensitivity, specificity]
print(cutoff_df)

     prob  Accuracy  sensitivity  specificity
0.0   0.0  0.381494     1.000000     0.000000
0.1   0.1  0.693588     0.981915     0.515748
0.2   0.2  0.723079     0.976241     0.566929
0.3   0.3  0.725649     0.957447     0.582677
0.4   0.4  0.835363     0.624113     0.965661
0.5   0.5  0.836851     0.620922     0.970035
0.6   0.6  0.836580     0.618085     0.971347
0.7   0.7  0.834145     0.609220     0.972878
0.8   0.8  0.834010     0.608865     0.972878
0.9   0.9  0.705357     0.230851     0.998031

# let's plot the accuracy, specificity, sensitivity 
cutoff_df.plot.line(x='prob',y=['Accuracy','sensitivity','specificity'])
plt.show()

y_train_pred_final['final_predicted'] = y_train_pred_final.converted_prob.map( lambda x: 1 if x > 0.4 else 0)

y_train_pred_final.head()

# Accuracy score
metrics.accuracy_score(y_train_pred_final.Converted, y_train_pred_final.final_predicted)

0.8353625541125541

# Confusion matrix
confusion_matrix2 = metrics.confusion_matrix(y_train_pred_final.Converted,y_train_pred_final.final_predicted)
confusion_matrix2

array([[4415,  157],
       [1060, 1760]], dtype=int64)

TP = confusion_matrix2[1,1]
FP = confusion_matrix2[0,1]
TN = confusion_matrix2[0,0]
FN = confusion_matrix2[1,0]

# Sensitivity
TP/float(TP+FN)

0.624113475177305

# specificity
TN/float(TN+FP)

0.965660542432196

# False positive rate
FP/float(TN+FP)

0.03433945756780402

# positive predictive rate
TP/float(TP+FP)

0.9181011997913406

# negative predictive value
TN/float(TN+FN)

0.806392694063927

Precision = TP/float(FP+TP)
Precision

0.9181011997913406

Recall = TP/float(TP+FN)
Recall

0.624113475177305

# Using sklearn utilities for the same
from sklearn.metrics import precision_score, recall_score, precision_recall_curve

precision = precision_score(y_train_pred_final.Converted,y_train_pred_final.final_predicted)
precision

0.9181011997913406

Recall = recall_score(y_train_pred_final.Converted,y_train_pred_final.final_predicted)
Recall

0.624113475177305

p, r, thresholds = precision_recall_curve(y_train_pred_final.Converted, y_train_pred_final.converted_prob)

plt.plot(thresholds, p[:-1], "g-")
plt.plot(thresholds, r[:-1], "r-")
plt.show()

# Scaling the continous variables in the test dataset
x_test[['TotalVisits','Total Time Spent on Website','Page Views Per Visit']]= scaler.transform(x_test[['TotalVisits',
                                                                                                             'Total Time Spent on Website',
                                                                                                             'Page Views Per Visit']])

x_test = x_test[col]
x_test.head()

# Add constant to the x_test variables
x_test_sm = sm.add_constant(x_test)
x_test_sm.shape

(1848, 13)

# Making predictions on test data
y_test_pred = res.predict(x_test_sm)
y_test_pred

4608    0.987269
7935    0.107967
4043    0.107967
7821    0.004608
856     0.358775
          ...   
7387    0.358775
3063    0.971569
603     0.004608
4210    0.358775
7352    0.004608
Length: 1848, dtype: float64

# Creating a dataframe "y_pred_1"
y_pred_1 = pd.DataFrame(y_test_pred)

# creating a dataframe "y_test_df"
y_test_df = pd.DataFrame(y_test)

# adding prospect id column
y_test_df['Prospect_ID'] = y_test_df.index

# Removing index for both dataframes to append them side by side 
y_pred_1.reset_index(drop=True, inplace=True)
y_test_df.reset_index(drop=True, inplace=True)

# Appending y_test_df and y_pred_1 datafromes 
y_pred_final = pd.concat([y_test_df,y_pred_1],axis=1)
y_pred_final.head()

# Renaming the column
y_pred_final = y_pred_final.rename(columns = {0: 'Converted_prob'})
y_pred_final

# Rearranging the columns
y_pred_final = y_pred_final.reindex(['Prospect_ID','Converted','Converted_prob'],axis=1)
y_pred_final.head()

# Mapping the converted prob to 0 or 1
y_pred_final['final_predicted'] = y_pred_final.Converted_prob.map(lambda x: 1 if x>0.38 else 0)

y_pred_final.head()

# calculating the lead scores (0-100)
y_pred_final['Lead score'] = (y_pred_final['Converted_prob']*100).round(0).astype(int)
y_pred_final

# Accuracy
metrics.accuracy_score(y_pred_final.Converted,y_pred_final.final_predicted)

0.8284632034632035

confusion3 = metrics.confusion_matrix(y_pred_final.Converted,y_pred_final.final_predicted)
confusion3

array([[1061,   46],
       [ 271,  470]], dtype=int64)

TP = confusion3[1,1] # true positive 
TN = confusion3[0,0] # true negatives
FP = confusion3[0,1] # false positives
FN = confusion3[1,0] # false negatives

# Let's see the sensitivity of our logistic regression model
TP / float(TP+FN)

0.6342780026990553

# Let us calculate specificity
TN / float(TN+FP)

0.9584462511291779

# Precision
Precision = TP/float(FP+TP)
Precision

0.9108527131782945

# Recall
Recall = TP/float(TP+FN)
Recall

0.6342780026990553

# Tagging the users as "very hot leads",'warm leads", "cold leads" based on their lead score
y_pred_final['Interpretation'] = y_pred_final['Lead score'].map(lambda x: 'Very Hot Leads' if x>80 else 'Warm leads' if (50<x<79) else 'Cold Leads' )
y_pred_final

import pandas as pd
import numpy as np

# get feature names + coefficients directly from statsmodels
coef_df = pd.DataFrame({
    'Feature': res.params.index,
    'Coefficient': res.params.values
})
coef_df = coef_df[coef_df['Feature'] != 'const']
# add absolute coefficient for ranking
coef_df['Abs_Coefficient'] = np.abs(coef_df['Coefficient'])

# sort by absolute coefficient
coef_df = coef_df.sort_values(by='Abs_Coefficient', ascending=False)

# top 3 variables
top_3 = coef_df.head(3)
print(top_3)

                                           Feature  Coefficient  \
5                          Tags_Closed by Horizzon     8.305587   
6                                Tags_Lost to EINS     7.486129   
10  Last Notable Activity_Had a Phone Conversation     3.799912   

    Abs_Coefficient  
5          8.305587  
6          7.486129  
10         3.799912

# Assuming 'res' is your fitted statsmodels Logit/GLM model
coefficients = res.params  

# Put into dataframe
coef_df = pd.DataFrame({
    'Feature': coefficients.index,
    'Coefficient': coefficients.values,
    'Abs_Coefficient': np.abs(coefficients.values)
})

# Keep only categorical (dummy) variables
dummy_vars = [col for col in coef_df['Feature'] if "_" in col]  # dummies usually have "_" in names
coef_df = coef_df[coef_df['Feature'].isin(dummy_vars)]

# Sort by absolute coefficient
coef_df = coef_df.sort_values(by='Abs_Coefficient', ascending=False)

# Get top 3
top3 = coef_df.head(3)
print(top3)

                                           Feature  Coefficient  \
5                          Tags_Closed by Horizzon     8.305587   
6                                Tags_Lost to EINS     7.486129   
10  Last Notable Activity_Had a Phone Conversation     3.799912   

    Abs_Coefficient  
5          8.305587  
6          7.486129  
10         3.799912

# Reducing decision cuttoff to 0.3
y_train_pred_final['final_predicted'] = y_train_pred_final.converted_prob.map( lambda x: 1 if x > 0.3 else 0)

y_train_pred_final.head()

# Accuracy score
metrics.accuracy_score(y_train_pred_final.Converted, y_train_pred_final.final_predicted)

0.7256493506493507

# confusion Matrix 
confusion_matrix2 = metrics.confusion_matrix(y_train_pred_final.Converted,y_train_pred_final.final_predicted)

TP = confusion_matrix2[1,1] # True positive
FP = confusion_matrix2[0,1] # False positive
TN = confusion_matrix2[0,0] # True negative
FN = confusion_matrix2[1,0] # False negative

Precision = TP/float(FP+TP)
Precision

0.5859375

Recall = TP/float(TP+FN)
Recall

0.9574468085106383

# Increasing decision cuttoff to 0.3
y_train_pred_final['final_predicted'] = y_train_pred_final.converted_prob.map( lambda x: 1 if x > 0.7 else 0)

y_train_pred_final.head()

# Accuracy score
metrics.accuracy_score(y_train_pred_final.Converted, y_train_pred_final.final_predicted)

0.8341450216450217

# confusion Matrix 
confusion_matrix3 = metrics.confusion_matrix(y_train_pred_final.Converted,y_train_pred_final.final_predicted)

TP = confusion_matrix3[1,1] # True positive
FP = confusion_matrix3[0,1] # False positive
TN = confusion_matrix3[0,0] # True negative
FN = confusion_matrix3[1,0] # False negative

Precision = TP/float(FP+TP)
Precision

0.9326818675352877

Recall = TP/float(TP+FN)
Recall

0.6092198581560284

	Lead Number	Converted	TotalVisits	Total Time Spent on Website	Page Views Per Visit	Asymmetrique Activity Score	Asymmetrique Profile Score
count	9240.000000	9240.000000	9103.000000	9240.000000	9103.000000	5022.000000	5022.000000
mean	617188.435606	0.385390	3.445238	487.698268	2.362820	14.306252	16.344883
std	23405.995698	0.486714	4.854853	548.021466	2.161418	1.386694	1.811395
min	579533.000000	0.000000	0.000000	0.000000	0.000000	7.000000	11.000000
25%	596484.500000	0.000000	1.000000	12.000000	1.000000	14.000000	15.000000
50%	615479.000000	0.000000	3.000000	248.000000	2.000000	14.000000	16.000000
75%	637387.250000	1.000000	5.000000	936.000000	3.000000	15.000000	18.000000
max	660737.000000	1.000000	251.000000	2272.000000	55.000000	18.000000	20.000000

	Lead Number	Converted	TotalVisits	Total Time Spent on Website	Page Views Per Visit
count	9240.000000	9240.000000	9240.000000	9240.000000	9240.000000
mean	617188.435606	0.385390	3.438636	487.698268	2.357440
std	23405.995698	0.486714	4.819024	548.021466	2.145781
min	579533.000000	0.000000	0.000000	0.000000	0.000000
25%	596484.500000	0.000000	1.000000	12.000000	1.000000
50%	615479.000000	0.000000	3.000000	248.000000	2.000000
75%	637387.250000	1.000000	5.000000	936.000000	3.000000
90%	650506.100000	1.000000	7.000000	1380.000000	5.000000
95%	655404.050000	1.000000	10.000000	1562.000000	6.000000
99%	659592.980000	1.000000	17.000000	1840.610000	9.000000
max	660737.000000	1.000000	251.000000	2272.000000	55.000000

Dep. Variable:	Converted	No. Observations:	7392
Model:	GLM	Df Residuals:	7245
Model Family:	Binomial	Df Model:	146
Link Function:	Logit	Scale:	1.0000
Method:	IRLS	Log-Likelihood:	nan
Date:	Fri, 26 Sep 2025	Deviance:	93393.
Time:	20:13:05	Pearson chi2:	4.57e+18
No. Iterations:	100	Pseudo R-squ. (CS):	nan
Covariance Type:	nonrobust

	coef	std err	z	P>\|z\|	[0.025	0.975]
const	1.304e+16	9.4e+07	1.39e+08	0.000	1.3e+16	1.3e+16
Do Not Email	-4.491e+14	4.14e+06	-1.09e+08	0.000	-4.49e+14	-4.49e+14
Do Not Call	7.764e+15	4.77e+07	1.63e+08	0.000	7.76e+15	7.76e+15
TotalVisits	8.281e+13	1.32e+06	6.29e+07	0.000	8.28e+13	8.28e+13
Total Time Spent on Website	4.253e+14	9.26e+05	4.6e+08	0.000	4.25e+14	4.25e+14
Page Views Per Visit	-9.822e+13	1.44e+06	-6.8e+07	0.000	-9.82e+13	-9.82e+13
Search	7.833e+14	2.11e+07	3.72e+07	0.000	7.83e+14	7.83e+14
Magazine	83.2603	8.93e-07	9.33e+07	0.000	83.260	83.260
Newspaper Article	2.107e+15	6.73e+07	3.13e+07	0.000	2.11e+15	2.11e+15
X Education Forums	-9.99e+15	1.08e+08	-9.27e+07	0.000	-9.99e+15	-9.99e+15
Newspaper	-7.93e+15	6.73e+07	-1.18e+08	0.000	-7.93e+15	-7.93e+15
Digital Advertisement	5.979e+14	3.91e+07	1.53e+07	0.000	5.98e+14	5.98e+14
Through Recommendations	1.285e+15	2.84e+07	4.52e+07	0.000	1.28e+15	1.28e+15
Receive More Updates About Our Courses	75.4163	1.06e-06	7.14e+07	0.000	75.416	75.416
Update me on Supply Chain Content	-368.3959	2.09e-06	-1.76e+08	0.000	-368.396	-368.396
Get updates on DM Content	172.3771	1.11e-06	1.56e+08	0.000	172.377	172.377
I agree to pay the amount through cheque	-124.1646	1.6e-06	-7.76e+07	0.000	-124.165	-124.165
A free copy of Mastering The Interview	-1.357e+14	2.52e+06	-5.38e+07	0.000	-1.36e+14	-1.36e+14
Lead Origin_Landing Page Submission	-1.867e+14	2.81e+06	-6.65e+07	0.000	-1.87e+14	-1.87e+14
Lead Origin_Lead Add Form	8.759e+14	1.3e+07	6.73e+07	0.000	8.76e+14	8.76e+14
Lead Origin_Lead Import	2.215e+15	4.93e+07	4.49e+07	0.000	2.22e+15	2.22e+15
Lead Origin_Quick Add Form	1.19e+16	6.79e+07	1.75e+08	0.000	1.19e+16	1.19e+16
Lead Source_Direct Traffic	-9.534e+14	4.2e+07	-2.27e+07	0.000	-9.53e+14	-9.53e+14
Lead Source_Facebook	-3.326e+15	6.47e+07	-5.14e+07	0.000	-3.33e+15	-3.33e+15
Lead Source_Google	-9.919e+14	4.2e+07	-2.36e+07	0.000	-9.92e+14	-9.92e+14
Lead Source_Live Chat	3.077e+14	6.21e+07	4.96e+06	0.000	3.08e+14	3.08e+14
Lead Source_NC_EDM	4.096e+15	7.92e+07	5.17e+07	0.000	4.1e+15	4.1e+15
Lead Source_Olark Chat	-8.88e+14	4.21e+07	-2.11e+07	0.000	-8.88e+14	-8.88e+14
Lead Source_Organic Search	-8.593e+14	4.2e+07	-2.04e+07	0.000	-8.59e+14	-8.59e+14
Lead Source_Pay per Click Ads	-3.839e+15	7.93e+07	-4.84e+07	0.000	-3.84e+15	-3.84e+15
Lead Source_Press_Release	-3.9298	3.8e-07	-1.03e+07	0.000	-3.930	-3.930
Lead Source_Reference	-6.961e+14	4.01e+07	-1.74e+07	0.000	-6.96e+14	-6.96e+14
Lead Source_Referral Sites	-7.789e+14	4.25e+07	-1.83e+07	0.000	-7.79e+14	-7.79e+14
Lead Source_Social Media	-1.16e+15	6.35e+07	-1.83e+07	0.000	-1.16e+15	-1.16e+15
Lead Source_WeLearn	5.764e+15	7.92e+07	7.27e+07	0.000	5.76e+15	5.76e+15
Lead Source_Welingak Website	-3.273e+14	4.06e+07	-8.07e+06	0.000	-3.27e+14	-3.27e+14
Lead Source_bing	-4.644e+14	5.17e+07	-8.99e+06	0.000	-4.64e+14	-4.64e+14
Lead Source_blog	-3.23e+15	7.93e+07	-4.08e+07	0.000	-3.23e+15	-3.23e+15
Lead Source_google	-2.075e+15	5.18e+07	-4e+07	0.000	-2.07e+15	-2.07e+15
Lead Source_testone	-3.733e+15	7.95e+07	-4.69e+07	0.000	-3.73e+15	-3.73e+15
Lead Source_welearnblog_Home	-362.5757	1.95e-06	-1.86e+08	0.000	-362.576	-362.576
Lead Source_youtubechannel	-1.07e+15	9.34e+07	-1.15e+07	0.000	-1.07e+15	-1.07e+15
Last Activity_Converted to Lead	-1.525e+15	2.44e+07	-6.25e+07	0.000	-1.53e+15	-1.53e+15
Last Activity_Email Bounced	-1.523e+15	2.48e+07	-6.13e+07	0.000	-1.52e+15	-1.52e+15
Last Activity_Email Link Clicked	-1.307e+15	2.52e+07	-5.19e+07	0.000	-1.31e+15	-1.31e+15
Last Activity_Email Marked Spam	-4.607e+15	4.15e+07	-1.11e+08	0.000	-4.61e+15	-4.61e+15
Last Activity_Email Opened	-1.398e+15	2.41e+07	-5.8e+07	0.000	-1.4e+15	-1.4e+15
Last Activity_Email Received	2.695e+16	7.14e+07	3.78e+08	0.000	2.7e+16	2.7e+16
Last Activity_Form Submitted on Website	-1.571e+15	2.51e+07	-6.26e+07	0.000	-1.57e+15	-1.57e+15
Last Activity_Had a Phone Conversation	-1.539e+15	3.03e+07	-5.08e+07	0.000	-1.54e+15	-1.54e+15
Last Activity_Olark Chat Conversation	-9.413e+14	2.42e+07	-3.89e+07	0.000	-9.41e+14	-9.41e+14
Last Activity_Page Visited on Website	-1.594e+15	2.44e+07	-6.52e+07	0.000	-1.59e+15	-1.59e+15
Last Activity_Resubscribed to emails	-60.4735	6.35e-07	-9.53e+07	0.000	-60.474	-60.474
Last Activity_SMS Sent	-1.263e+15	2.42e+07	-5.21e+07	0.000	-1.26e+15	-1.26e+15
Last Activity_Unreachable	-1.415e+15	2.61e+07	-5.43e+07	0.000	-1.42e+15	-1.42e+15
Last Activity_Unsubscribed	-2.456e+15	3.11e+07	-7.91e+07	0.000	-2.46e+15	-2.46e+15
Last Activity_View in browser link Clicked	-1.267e+15	4.14e+07	-3.06e+07	0.000	-1.27e+15	-1.27e+15
Last Activity_Visited Booth in Tradeshow	1.027e+15	7.21e+07	1.42e+07	0.000	1.03e+15	1.03e+15
Country_Australia	7.261e+12	5.27e+07	1.38e+05	0.000	7.26e+12	7.26e+12
Country_Bahrain	1.462e+15	5.63e+07	2.59e+07	0.000	1.46e+15	1.46e+15
Country_Bangladesh	6.949e+14	6.73e+07	1.03e+07	0.000	6.95e+14	6.95e+14
Country_Belgium	-6.948e+15	6.74e+07	-1.03e+08	0.000	-6.95e+15	-6.95e+15
Country_Canada	-3.783e+14	6.74e+07	-5.61e+06	0.000	-3.78e+14	-3.78e+14
Country_China	-3.181e+15	6.72e+07	-4.73e+07	0.000	-3.18e+15	-3.18e+15
Country_Denmark	1.59e+15	8.24e+07	1.93e+07	0.000	1.59e+15	1.59e+15
Country_France	3.332e+14	5.64e+07	5.91e+06	0.000	3.33e+14	3.33e+14
Country_Germany	6.687e+14	5.83e+07	1.15e+07	0.000	6.69e+14	6.69e+14
Country_Ghana	-2.716e+15	8.23e+07	-3.3e+07	0.000	-2.72e+15	-2.72e+15
Country_Hong Kong	1.192e+15	5.49e+07	2.17e+07	0.000	1.19e+15	1.19e+15
Country_India	2.573e+14	4.75e+07	5.41e+06	0.000	2.57e+14	2.57e+14
Country_Indonesia	27.5231	3.43e-07	8.02e+07	0.000	27.523	27.523
Country_Italy	-7.526e+15	6.76e+07	-1.11e+08	0.000	-7.53e+15	-7.53e+15
Country_Kenya	-4.678e+15	8.24e+07	-5.68e+07	0.000	-4.68e+15	-4.68e+15
Country_Kuwait	3.061e+14	6.15e+07	4.98e+06	0.000	3.06e+14	3.06e+14
Country_Liberia	-2.553e+15	8.26e+07	-3.09e+07	0.000	-2.55e+15	-2.55e+15
Country_Malaysia	-5.8e+14	8.28e+07	-7.01e+06	0.000	-5.8e+14	-5.8e+14
Country_Netherlands	2.155e+15	8.24e+07	2.62e+07	0.000	2.16e+15	2.16e+15
Country_Nigeria	-1.658e+15	6.15e+07	-2.7e+07	0.000	-1.66e+15	-1.66e+15
Country_Oman	5.073e+14	6.15e+07	8.25e+06	0.000	5.07e+14	5.07e+14
Country_Philippines	-2.038e+15	8.24e+07	-2.47e+07	0.000	-2.04e+15	-2.04e+15
Country_Qatar	-9.695e+12	5.26e+07	-1.84e+05	0.000	-9.7e+12	-9.7e+12
Country_Russia	17.7277	1.21e-07	1.46e+08	0.000	17.728	17.728
Country_Saudi Arabia	3.506e+14	5.03e+07	6.97e+06	0.000	3.51e+14	3.51e+14
Country_Singapore	7.518e+14	5e+07	1.5e+07	0.000	7.52e+14	7.52e+14
Country_South Africa	-7.621e+15	6.74e+07	-1.13e+08	0.000	-7.62e+15	-7.62e+15
Country_Sri Lanka	-4.199e+15	8.26e+07	-5.08e+07	0.000	-4.2e+15	-4.2e+15
Country_Sweden	-2.908e+14	6.13e+07	-4.74e+06	0.000	-2.91e+14	-2.91e+14
Country_Switzerland	-3.688e+15	8.24e+07	-4.48e+07	0.000	-3.69e+15	-3.69e+15
Country_Tanzania	-6.373e+15	8.26e+07	-7.71e+07	0.000	-6.37e+15	-6.37e+15
Country_Uganda	-8.12e+15	6.73e+07	-1.21e+08	0.000	-8.12e+15	-8.12e+15
Country_United Arab Emirates	4.267e+14	4.87e+07	8.77e+06	0.000	4.27e+14	4.27e+14
Country_United Kingdom	5.658e+14	5.14e+07	1.1e+07	0.000	5.66e+14	5.66e+14
Country_United States	5.65e+14	4.84e+07	1.17e+07	0.000	5.65e+14	5.65e+14
Country_Vietnam	5.7824	1.07e-07	5.39e+07	0.000	5.782	5.782
Country_unknown	1.154e+15	6.14e+07	1.88e+07	0.000	1.15e+15	1.15e+15
Specialization_Business Administration	-2.254e+13	5.57e+06	-4.05e+06	0.000	-2.25e+13	-2.25e+13
Specialization_E-Business	-4.093e+14	1.07e+07	-3.81e+07	0.000	-4.09e+14	-4.09e+14
Specialization_E-COMMERCE	-3.287e+14	8.23e+06	-3.99e+07	0.000	-3.29e+14	-3.29e+14
Specialization_Finance Management	-5.233e+14	4.39e+06	-1.19e+08	0.000	-5.23e+14	-5.23e+14
Specialization_Healthcare Management	-1.034e+14	7.29e+06	-1.42e+07	0.000	-1.03e+14	-1.03e+14
Specialization_Hospitality Management	-3.198e+14	8.18e+06	-3.91e+07	0.000	-3.2e+14	-3.2e+14
Specialization_Human Resource Management	-1.586e+14	4.79e+06	-3.31e+07	0.000	-1.59e+14	-1.59e+14
Specialization_IT Projects Management	-5.802e+13	5.63e+06	-1.03e+07	0.000	-5.8e+13	-5.8e+13
Specialization_International Business	-3.956e+14	6.86e+06	-5.77e+07	0.000	-3.96e+14	-3.96e+14
Specialization_Marketing Management	-5.792e+13	4.78e+06	-1.21e+07	0.000	-5.79e+13	-5.79e+13
Specialization_Media and Advertising	-1.647e+14	6.62e+06	-2.49e+07	0.000	-1.65e+14	-1.65e+14
Specialization_Operations Management	-9.508e+13	5.25e+06	-1.81e+07	0.000	-9.51e+13	-9.51e+13
Specialization_Retail Management	-3.182e+14	8.47e+06	-3.76e+07	0.000	-3.18e+14	-3.18e+14
Specialization_Rural and Agribusiness	-1.929e+14	9.88e+06	-1.95e+07	0.000	-1.93e+14	-1.93e+14
Specialization_Services Excellence	-3.007e+14	1.23e+07	-2.45e+07	0.000	-3.01e+14	-3.01e+14
Specialization_Supply Chain Management	-2.245e+14	5.76e+06	-3.89e+07	0.000	-2.24e+14	-2.24e+14
Specialization_Travel and Tourism	-1.29e+14	6.71e+06	-1.92e+07	0.000	-1.29e+14	-1.29e+14
What is your current occupation_Housewife	3.677e+15	2.14e+07	1.72e+08	0.000	3.68e+15	3.68e+15
What is your current occupation_Other	4.496e+14	2.14e+07	2.1e+07	0.000	4.5e+14	4.5e+14
What is your current occupation_Student	4.869e+14	5.72e+06	8.51e+07	0.000	4.87e+14	4.87e+14
What matters most to you in choosing a course_Flexibility & Convenience	1.665e+15	6.75e+07	2.47e+07	0.000	1.66e+15	1.66e+15
What matters most to you in choosing a course_Other	-3.438e+15	6.77e+07	-5.08e+07	0.000	-3.44e+15	-3.44e+15
Tags_Busy	1.294e+15	6.9e+06	1.88e+08	0.000	1.29e+15	1.29e+15
Tags_Closed by Horizzon	2.339e+15	5.91e+06	3.96e+08	0.000	2.34e+15	2.34e+15
Tags_Diploma holder (Not Eligible)	-2.987e+15	1.08e+07	-2.76e+08	0.000	-2.99e+15	-2.99e+15
Tags_Graduation in progress	2.658e+14	8.2e+06	3.24e+07	0.000	2.66e+14	2.66e+14
Tags_In confusion whether part time or DLP	1.269e+15	3.38e+07	3.75e+07	0.000	1.27e+15	1.27e+15
Tags_Interested in full time MBA	1.02e+14	8.07e+06	1.26e+07	0.000	1.02e+14	1.02e+14
Tags_Interested in Next batch	4.409e+15	3.04e+07	1.45e+08	0.000	4.41e+15	4.41e+15
Tags_Interested in other courses	7.297e+12	5e+06	1.46e+06	0.000	7.3e+12	7.3e+12
Tags_Lateral student	6.567e+15	6.73e+07	9.75e+07	0.000	6.57e+15	6.57e+15
Tags_Lost to EINS	2.374e+15	6.77e+06	3.5e+08	0.000	2.37e+15	2.37e+15
Tags_Lost to Others	-2.498e+14	2.8e+07	-8.92e+06	0.000	-2.5e+14	-2.5e+14
Tags_Not doing further education	-5.533e+14	7.38e+06	-7.49e+07	0.000	-5.53e+14	-5.53e+14
Tags_Recognition issue (DEC approval)	7.9648	3.36e-08	2.37e+08	0.000	7.965	7.965
Tags_Ringing	-3.555e+14	4.35e+06	-8.16e+07	0.000	-3.55e+14	-3.55e+14
Tags_Shall take in the next coming month	-3.932e+15	6.73e+07	-5.84e+07	0.000	-3.93e+15	-3.93e+15
Tags_Still Thinking	-1.798e+15	3.04e+07	-5.92e+07	0.000	-1.8e+15	-1.8e+15
Tags_University not recognized	-1.259e+15	4.79e+07	-2.63e+07	0.000	-1.26e+15	-1.26e+15
Tags_Want to take admission but has financial problems	5.213e+14	2.87e+07	1.82e+07	0.000	5.21e+14	5.21e+14
Tags_Will revert after reading the email	7.926e+14	3.87e+06	2.05e+08	0.000	7.93e+14	7.93e+14
Tags_in touch with EINS	8.055e+14	2.06e+07	3.91e+07	0.000	8.06e+14	8.06e+14
Tags_invalid number	-4.604e+14	9.19e+06	-5.01e+07	0.000	-4.6e+14	-4.6e+14
Tags_number not provided	-8.383e+14	1.44e+07	-5.82e+07	0.000	-8.38e+14	-8.38e+14
Tags_opp hangup	2.122e+14	1.25e+07	1.69e+07	0.000	2.12e+14	2.12e+14
Tags_switched off	-8.503e+14	6.19e+06	-1.37e+08	0.000	-8.5e+14	-8.5e+14
Tags_wrong number given	-4.881e+14	1.12e+07	-4.35e+07	0.000	-4.88e+14	-4.88e+14
City_Other Cities	8.205e+13	3.22e+06	2.55e+07	0.000	8.21e+13	8.21e+13
City_Other Cities of Maharashtra	4.574e+13	3.81e+06	1.2e+07	0.000	4.57e+13	4.57e+13
City_Other Metro Cities	2.513e+13	4.21e+06	5.96e+06	0.000	2.51e+13	2.51e+13
City_Thane & Outskirts	1.423e+14	3e+06	4.74e+07	0.000	1.42e+14	1.42e+14
City_Tier II Cities	2.168e+14	9.03e+06	2.4e+07	0.000	2.17e+14	2.17e+14
Last Notable Activity_Email Bounced	-1.124e+16	7.41e+07	-1.52e+08	0.000	-1.12e+16	-1.12e+16
Last Notable Activity_Email Link Clicked	-1.182e+16	7.39e+07	-1.6e+08	0.000	-1.18e+16	-1.18e+16
Last Notable Activity_Email Marked Spam	-4.607e+15	4.15e+07	-1.11e+08	0.000	-4.61e+15	-4.61e+15
Last Notable Activity_Email Opened	-1.21e+16	7.33e+07	-1.65e+08	0.000	-1.21e+16	-1.21e+16
Last Notable Activity_Email Received	-3.35e+16	1.2e+08	-2.79e+08	0.000	-3.35e+16	-3.35e+16
Last Notable Activity_Form Submitted on Website	0	0	nan	nan	0	0
Last Notable Activity_Had a Phone Conversation	-1.088e+16	7.79e+07	-1.4e+08	0.000	-1.09e+16	-1.09e+16
Last Notable Activity_Modified	-1.186e+16	7.33e+07	-1.62e+08	0.000	-1.19e+16	-1.19e+16
Last Notable Activity_Olark Chat Conversation	-1.227e+16	7.35e+07	-1.67e+08	0.000	-1.23e+16	-1.23e+16
Last Notable Activity_Page Visited on Website	-1.154e+16	7.35e+07	-1.57e+08	0.000	-1.15e+16	-1.15e+16
Last Notable Activity_Resubscribed to emails	0	0	nan	nan	0	0
Last Notable Activity_SMS Sent	-1.076e+16	7.34e+07	-1.47e+08	0.000	-1.08e+16	-1.08e+16
Last Notable Activity_Unreachable	-1.101e+16	7.5e+07	-1.47e+08	0.000	-1.1e+16	-1.1e+16
Last Notable Activity_Unsubscribed	-1.05e+16	7.67e+07	-1.37e+08	0.000	-1.05e+16	-1.05e+16
Last Notable Activity_View in browser link Clicked	-1.141e+16	1.05e+08	-1.09e+08	0.000	-1.14e+16	-1.14e+16

Dep. Variable:	Converted	No. Observations:	7392
Model:	GLM	Df Residuals:	7376
Model Family:	Binomial	Df Model:	15
Link Function:	Logit	Scale:	1.0000
Method:	IRLS	Log-Likelihood:	-2613.0
Date:	Fri, 26 Sep 2025	Deviance:	5225.9
Time:	20:13:12	Pearson chi2:	1.25e+04
No. Iterations:	23	Pseudo R-squ. (CS):	0.4635
Covariance Type:	nonrobust

	Prospect ID	Lead Number	Lead Origin	Lead Source	Do Not Email	Do Not Call	Converted	TotalVisits	Total Time Spent on Website	Page Views Per Visit	...	Get updates on DM Content	Lead Profile	City	Asymmetrique Activity Index	Asymmetrique Profile Index	Asymmetrique Activity Score	Asymmetrique Profile Score	I agree to pay the amount through cheque	A free copy of Mastering The Interview	Last Notable Activity
0	7927b2df-8bba-4d29-b9a2-b6e0beafe620	660737	API	Olark Chat	No	No	0	0.0	0	0.0	...	No	Select	Select	02.Medium	02.Medium	15.0	15.0	No	No	Modified
1	2a272436-5132-4136-86fa-dcc88c88f482	660728	API	Organic Search	No	No	0	5.0	674	2.5	...	No	Select	Select	02.Medium	02.Medium	15.0	15.0	No	No	Email Opened
2	8cc8c611-a219-4f35-ad23-fdfd2656bd8a	660727	Landing Page Submission	Direct Traffic	No	No	1	2.0	1532	2.0	...	No	Potential Lead	Mumbai	02.Medium	01.High	14.0	20.0	No	Yes	Email Opened
3	0cc2df48-7cf4-4e39-9de9-19797f9b38cc	660719	Landing Page Submission	Direct Traffic	No	No	0	1.0	305	1.0	...	No	Select	Mumbai	02.Medium	01.High	13.0	17.0	No	No	Modified
4	3256f628-e534-4826-9d63-4a8b88782852	660681	Landing Page Submission	Google	No	No	1	2.0	1428	1.0	...	No	Select	Mumbai	02.Medium	01.High	15.0	18.0	No	No	Modified

	Prospect ID	Lead Number	Lead Origin	Lead Source	Do Not Email	Do Not Call	Converted	TotalVisits	Total Time Spent on Website	Page Views Per Visit	...	Digital Advertisement	Through Recommendations	Receive More Updates About Our Courses	Tags	Update me on Supply Chain Content	Get updates on DM Content	City	I agree to pay the amount through cheque	A free copy of Mastering The Interview	Last Notable Activity
0	7927b2df-8bba-4d29-b9a2-b6e0beafe620	660737.0	API	Olark Chat	No	No	0.0	0.0	0.0	0.0	...	No	No	No	Interested in other courses	No	No	Mumbai	No	No	Modified
1	2a272436-5132-4136-86fa-dcc88c88f482	660728.0	API	Organic Search	No	No	0.0	5.0	674.0	2.5	...	No	No	No	Ringing	No	No	Mumbai	No	No	Email Opened
2	8cc8c611-a219-4f35-ad23-fdfd2656bd8a	660727.0	Landing Page Submission	Direct Traffic	No	No	1.0	2.0	1532.0	2.0	...	No	No	No	Will revert after reading the email	No	No	Mumbai	No	Yes	Email Opened
3	0cc2df48-7cf4-4e39-9de9-19797f9b38cc	660719.0	Landing Page Submission	Direct Traffic	No	No	0.0	1.0	305.0	1.0	...	No	No	No	Ringing	No	No	Mumbai	No	No	Modified
4	3256f628-e534-4826-9d63-4a8b88782852	660681.0	Landing Page Submission	Google	No	No	1.0	2.0	1428.0	1.0	...	No	No	No	Will revert after reading the email	No	No	Mumbai	No	No	Modified

	Do Not Email	TotalVisits	Total Time Spent on Website	Page Views Per Visit	...	Last Notable Activity_Modified	Last Notable Activity_SMS Sent
6487	1	-0.433198	-0.454165	-0.149720	...	0	1
4759	0	-1.129102	-0.889097	-1.272692	...	1	0
4368	0	-0.085246	-0.168456	0.411765	...	0	0
1467	0	0.262706	0.737805	0.973251	...	0	1
5517	0	-0.433198	-0.628866	-0.149720	...	1	0

	Converted	converted_prob	Prospect ID
0	0.0	0.015654	6487
1	0.0	0.357183	4759
2	0.0	0.024448	4368
3	1.0	0.901694	1467
4	0.0	0.024448	5517
...	...	...	...
7387	0.0	0.004408	5734
7388	0.0	0.357183	5191
7389	1.0	0.990350	5390
7390	0.0	0.004408	860
7391	1.0	0.357183	7270

	Features	VIF
1	Lead Origin_Lead Add Form	1.49
13	Last Notable Activity_SMS Sent	1.44
7	Tags_Will revert after reading the email	1.37
2	Lead Source_Welingak Website	1.22
4	Tags_Closed by Horizzon	1.16
0	Do Not Email	1.14
6	Tags_Ringing	1.09
14	Last Notable Activity_Unsubscribed	1.06
3	Tags_Busy	1.03
10	Tags_switched off	1.03
8	Tags_invalid number	1.01
9	Tags_number not provided	1.01
11	Tags_wrong number given	1.01
5	Tags_Lost to EINS	1.00
12	Last Notable Activity_Had a Phone Conversation	1.00

	Converted	converted_prob	Prospect ID	Predicted
0	0.0	0.015583	6487	0
1	0.0	0.357612	4759	0
2	0.0	0.022982	4368	0
3	1.0	0.900540	1467	1
4	0.0	0.022982	5517	0
...	...	...	...	...
7387	0.0	0.004461	5734	0
7388	0.0	0.357612	5191	0
7389	1.0	0.990270	5390	1
7390	0.0	0.004461	860	0
7391	1.0	0.357612	7270	0

	Converted	converted_prob	Prospect ID	Predicted
0	0.0	0.015403	6487	0
1	0.0	0.358201	4759	0
2	0.0	0.021078	4368	0
3	1.0	0.899118	1467	1
4	0.0	0.021078	5517	0

	Converted	converted_prob	Prospect ID	Predicted
0	0.0	0.015408	6487	0
1	0.0	0.358775	4759	0
2	0.0	0.018804	4368	0
3	1.0	0.897375	1467	1
4	0.0	0.018804	5517	0

	Do Not Email	Tags_Closed by Horizzon	Tags_Ringing	Tags_Will revert after reading the email
4608	0	1	0	0
7935	1	0	0	1
4043	1	0	0	1
7821	0	0	1	0
856	0	0	0	1

	Converted	Prospect_ID	0
0	1.0	4608	0.987269
1	0.0	7935	0.107967
2	0.0	4043	0.107967
3	0.0	7821	0.004608
4	0.0	856	0.358775

	coef	std err	z	P>\|z\|	[0.025	0.975]
const	-3.6865	0.176	-20.980	0.000	-4.031	-3.342
Do Not Email	-1.5251	0.170	-8.952	0.000	-1.859	-1.191
Lead Origin_Lead Add Form	2.4149	0.209	11.537	0.000	2.005	2.825
Lead Source_Welingak Website	1.9829	1.034	1.917	0.055	-0.045	4.010
Tags_Busy	2.7654	0.273	10.141	0.000	2.231	3.300
Tags_Closed by Horizzon	8.0383	0.734	10.959	0.000	6.601	9.476
Tags_Lost to EINS	7.2176	0.539	13.394	0.000	6.161	8.274
Tags_Ringing	-1.7335	0.281	-6.170	0.000	-2.284	-1.183
Tags_Will revert after reading the email	3.0988	0.178	17.372	0.000	2.749	3.448
Tags_invalid number	-2.0582	1.040	-1.980	0.048	-4.096	-0.021
Tags_number not provided	-22.0940	2.21e+04	-0.001	0.999	-4.33e+04	4.33e+04
Tags_switched off	-2.5735	0.741	-3.472	0.001	-4.026	-1.121
Tags_wrong number given	-21.9906	1.73e+04	-0.001	0.999	-3.39e+04	3.38e+04
Last Notable Activity_Had a Phone Conversation	3.7106	1.169	3.175	0.001	1.420	6.001
Last Notable Activity_SMS Sent	2.8038	0.102	27.602	0.000	2.605	3.003
Last Notable Activity_Unsubscribed	2.1135	0.507	4.171	0.000	1.120	3.107

	coef	std err	z	P>\|z\|	[0.025	0.975]
const	-3.7498	0.176	-21.259	0.000	-4.095	-3.404
Do Not Email	-1.5271	0.170	-8.982	0.000	-1.860	-1.194
Lead Origin_Lead Add Form	2.4195	0.210	11.537	0.000	2.008	2.831
Lead Source_Welingak Website	1.9780	1.035	1.912	0.056	-0.050	4.006
Tags_Busy	2.8342	0.272	10.407	0.000	2.300	3.368
Tags_Closed by Horizzon	8.1015	0.734	11.042	0.000	6.663	9.539
Tags_Lost to EINS	7.2810	0.539	13.507	0.000	6.224	8.338
Tags_Ringing	-1.6580	0.280	-5.919	0.000	-2.207	-1.109
Tags_Will revert after reading the email	3.1640	0.179	17.698	0.000	2.814	3.514
Tags_invalid number	-1.9821	1.039	-1.907	0.057	-4.019	0.055
Tags_switched off	-2.4968	0.741	-3.371	0.001	-3.949	-1.045
Tags_wrong number given	-21.9152	1.73e+04	-0.001	0.999	-3.39e+04	3.39e+04
Last Notable Activity_Had a Phone Conversation	3.7311	1.176	3.171	0.002	1.425	6.037
Last Notable Activity_SMS Sent	2.7890	0.101	27.670	0.000	2.591	2.987
Last Notable Activity_Unsubscribed	2.1155	0.507	4.172	0.000	1.122	3.109

	coef	std err	z	P>\|z\|	[0.025	0.975]
const	-3.8382	0.177	-21.672	0.000	-4.185	-3.491
Do Not Email	-1.5355	0.169	-9.065	0.000	-1.868	-1.204
Lead Origin_Lead Add Form	2.4263	0.210	11.537	0.000	2.014	2.838
Lead Source_Welingak Website	1.9712	1.035	1.905	0.057	-0.057	3.999
Tags_Busy	2.9297	0.272	10.779	0.000	2.397	3.462
Tags_Closed by Horizzon	8.1899	0.734	11.160	0.000	6.752	9.628
Tags_Lost to EINS	7.3701	0.539	13.666	0.000	6.313	8.427
Tags_Ringing	-1.5545	0.279	-5.572	0.000	-2.101	-1.008
Tags_Will revert after reading the email	3.2550	0.179	18.173	0.000	2.904	3.606
Tags_invalid number	-1.8778	1.039	-1.807	0.071	-3.914	0.158
Tags_switched off	-2.3915	0.740	-3.231	0.001	-3.842	-0.941
Last Notable Activity_Had a Phone Conversation	3.7602	1.188	3.165	0.002	1.432	6.089
Last Notable Activity_SMS Sent	2.7706	0.100	27.736	0.000	2.575	2.966
Last Notable Activity_Unsubscribed	2.1239	0.508	4.185	0.000	1.129	3.119

	coef	std err	z	P>\|z\|	[0.025	0.975]
const	-3.9547	0.176	-22.453	0.000	-4.300	-3.609
Do Not Email	-1.5310	0.169	-9.052	0.000	-1.863	-1.199
Lead Origin_Lead Add Form	2.4335	0.211	11.533	0.000	2.020	2.847
Lead Source_Welingak Website	1.9628	1.035	1.897	0.058	-0.065	3.991
Tags_Busy	3.0538	0.270	11.316	0.000	2.525	3.583
Tags_Closed by Horizzon	8.3056	0.734	11.321	0.000	6.868	9.744
Tags_Lost to EINS	7.4861	0.539	13.889	0.000	6.430	8.543
Tags_Ringing	-1.4207	0.276	-5.144	0.000	-1.962	-0.879
Tags_Will revert after reading the email	3.3740	0.178	18.988	0.000	3.026	3.722
Tags_switched off	-2.2562	0.739	-3.053	0.002	-3.705	-0.808
Last Notable Activity_Had a Phone Conversation	3.7999	1.204	3.156	0.002	1.440	6.160
Last Notable Activity_SMS Sent	2.7491	0.099	27.823	0.000	2.555	2.943
Last Notable Activity_Unsubscribed	2.1073	0.506	4.168	0.000	1.116	3.098

	Prospect_ID	Converted	Converted_prob	final_predicted	Lead score	Interpretation
0	4608	1.0	0.987269	1	99	Very Hot Leads
1	7935	0.0	0.107967	0	11	Cold Leads
2	4043	0.0	0.107967	0	11	Cold Leads
3	7821	0.0	0.004608	0	0	Cold Leads
4	856	0.0	0.358775	0	36	Cold Leads
...	...	...	...	...	...	...
1843	7387	1.0	0.358775	0	36	Cold Leads
1844	3063	1.0	0.971569	1	97	Very Hot Leads
1845	603	0.0	0.004608	0	0	Cold Leads
1846	4210	1.0	0.358775	0	36	Cold Leads
1847	7352	0.0	0.004608	0	0	Cold Leads

LEAD SCORE ANALYSIS¶

Case Study - Summary¶

1. Company:¶

2. Current Situation:¶

3. Business Goal:¶

4. Data Given:¶

5. Expected Results:¶

Step-1: Reading the Data¶

Step-2: Understanding Data¶

Step-3: Data Preparation¶

1. Null Value Handling¶

Hence the null values in the data are removed. Therefore there are no null values in the data¶

2. Outliers Handling¶

3. Identify numerical and categorical columns in the dataset¶

Converting some binary variables (Yes/No) to 0/1¶

For categorical variables with multiple levels, create dummy features (one-hot encoded)¶

Checking the conversion rate¶

Step-4 : Looking at correlations¶

Step-5: Split data into input and target variables¶

Step-6 : Feature Scaling¶

Few Insights on the Variables affecting lead conversion¶

Step -7: Model Building¶

Feature selection by using RFE¶

Assessing the model with statsmodel¶

Creating a dataframe with the actual converted value and the predicted probabilities¶

Creating new column 'predicted' with 1 if Churn_Prob > 0.5 else 0¶

Confusion matrix¶

Checking VIFs¶

Let's now check the VIFs again¶

All variables have a good value of VIF. So we need not drop any more variables and we can proceed with making predictions using this model only¶

Metrics beyond Accuracy¶

Step-9 : Plotting the ROC curve¶

Step-10: Finding the optimal cutoff point¶

From the curve above, 0.4 is the optimum point to take it as a cutoff probability.¶

Precision and Recall¶

Precision and recall tradeoff¶

Step - 11: Making predictions on the test data (Model Evaluation)¶

Generating lead scores (0-100)¶

Answering Business Questions¶

Question -1:¶

Question -2:¶

Question -3:¶

By reducing the Decision Cutoff (Threshold) we can achieve this¶

Question-4:¶